Search results

Any plan to support Qwen3-VL dense model?

Status: Open.

#34 In NVlabs/Long-RL;

· RuixiangZhao opened on Oct 24, 2025

How to train with multi images?

Status: Open.

#33 In NVlabs/Long-RL;

· goodstudent9 opened on Sep 30, 2025

If this structure fit for NVILA-LITE-8B

Status: Open.

#32 In NVlabs/Long-RL;

· goodstudent9 opened on Sep 29, 2025

problem when run example/new_support/qwen2.5_omni_3b_grpo.sh

Status: Open.

#31 In NVlabs/Long-RL;

· wyclike opened on Sep 23, 2025

Why <code>log_probs.unsqueeze</code> in <code>compute_policy_loss</code>?

Status: Open.

#30 In NVlabs/Long-RL;

· Disapole-Xiao opened on Sep 18, 2025

The weights synchronizing of rollout

Status: Open.

#29 In NVlabs/Long-RL;

· coderwayne3025 opened on Sep 18, 2025

Can not run the Qwen-2.5-VL-7B model

Status: Open.

#28 In NVlabs/Long-RL;

· 64327069 opened on Sep 10, 2025

Wan text encoder is loaded on CPU?

Status: Open.

#27 In NVlabs/Long-RL;

· abcdvzz opened on Sep 8, 2025

Why is <code>ground_truth</code> not included in <code>non_tensor_batch_keys</code> in <code>ray_trainer.py</code>?

Status: Open.

#26 In NVlabs/Long-RL;

· rphlstck opened on Sep 2, 2025

Request for problem_type of training data

Status: Open.

#23 In NVlabs/Long-RL;

· vvukimy opened on Aug 20, 2025

The config of micro_batch_size_per_device_for_update and micro_batch_size_per_device_for_experience

Status: Open.

#21 In NVlabs/Long-RL;

· coderwayne3025 opened on Aug 18, 2025

Using Slurm For Training, World Size Settings

Status: Open.

#20 In NVlabs/Long-RL;

· Aurorana opened on Aug 13, 2025